Profiling Not-For-Profit Organisations with Machine Learning: The Role of Tax, Compensation, Acquisitions and Other Characteristics

Not-for-profit organisations (NPOs) are provided with support through tax reliefs and further funding in exchange for creating social benefit. They are, however, often accused of abusing and misallocating resources. Informed by configurational approaches, we apply a visual, non-parametric machine learning methodology to some of the largest UK annual financial micro-level datasets. The study profiles UK NPOs by establishing how their sectorial, growth and employment configurations differ from other organisations. The findings show that NPOs are more stable and more likely to become high growth firms. They have higher average employment than other organisations and most spend significantly more on employment-related costs and acquisitions than comparable organisations. The findings also show the interrelationships between lower tax paid and higher employment costs. The findings raise questions on the targeting and oversight of tax reductions, which are likely to be at least partly distributed through the inflated expenditure of larger NPOs.